A two stage neural network for choosing optimal ejection parameters in low altitude seat ejection based on novel injury parameter

Aircraft ejection systems are lifesaving devices used, in case of emergencies for military pilots. As per current literature the success rate of ejections happening in low altitudes (below 150 m) is 51.2%. The success rate of an ejection depends on various parameters such as height for parachute full deployment, ejection posture, spinal alignment, restraints etc. This paper proposes a multi objective optimization based on a novel injury parameter and the dynamic conditions of the aircraft prevailing during the time of emergency to choose the optimal impulse required for safe ejection. This injury parameter is capable of modelling the spine in different postures. A non-linear seat restraint interaction is also modelled to include the effect of varying slackness of seat restraints there by optimizing force applied for the different physical conditions of the pilot, which hasn’t been reported in literature yet. The Pareto solutions obtained thereof indicates the dangers associated with posture during the ejection process. The optimal solutions after decision making are used to train a two stage Artificial Neural Network architecture to choose optimal parameters for different ejection scenarios. The ANN proposed was seen to give good results which were along the generated Pareto fronts for different simulated aircraft conditions instantaneously.